Photoswitches in liquid speed detection

[ferragiov]

Hi,
I am in need to measure the velocity of a liquid running inside a plastic tube, so I set up a system of photoswitches and related control circuits. However, a wird thing is happening.
When there is no obstacle (just air) between the LED and the phototransistor, there is a current of about 2.25mA flowing in the transistor. When i place the empty plastic tube, the current falls to 1.5mA due to the loss of light (the tube is semi-transparent). So far so good, but when the liquid flows in front of the slit, the current rises to about 3mA, implying that the tube+liquid system conveys more light to the transistor than even the unobstructed photoswitch. Anybody know what could be going on? I tried several liquids of different density and color (just to try: water, red water, blue water, oil ecc) and there was absolutely no difference. The tube covers the slit completely, so this is not an effect of partial slit obscuring. Also, I tried in a completely dark environment, so it is not due to weird refraction of external light. I am using a Omron EE-SX1070 switch.

Thanks in advance,
Giovanni

 

[benta]

Could it be that tube+liquid work as a lens that focuses the light on the switch? That would account for it.

Benta.

 

[ferragiov]

I have working in the dark, so it is not due to external light focusing.It might be that tube+liquid focuses LED light to the transistor, but then I would assume the entity of this phenomenom would somehow change with the type of liquid. However, this is not the case, as it reacts the exact same with with different liquids (tried water, dyed water and oil). I also tried a tube of a different plastic, but the issue remains.

 

[VEBill]

It seems like you have a very good signal (1.5 -> 3 mA), just that the polarity isn't what you expected.

Also, this is to detect the speed of the air/liquid interface using two widely separated sensors, not the speed of the liquid when the tube is full, right?

 

[benta]

That's what I meant, focusing the LED on the transistor.
But given that this is an IR device, you don't really know how your dye works.

Benta.

 

[ferragiov]

It was originally intended to do so, to measure the air/liquid interface speed with two sensors. However, since the liquid is very slow (um/sec), and the sensor provides data for Collector current vs. Displacement with respect to optical axis, I was hoping to take the slope of the current-time curve with a simple derivator and to extract speed with a single measure. However this weird effect is messing things up. It could still be used with comparators to measure average speed, yes

 

[ferragiov]

@benta: yes, being IR I really don't know the effect of the dye, but I tried with different liquid (with different density) and still I don't measure any difference.

 

[IRstuff]

Density is not the same as refractive index, which is what will determine whether focussing occurs. And so long as the base solvent is relatively transparent to the naked eye, the indices of refraction will be roughly in the same ballpark. Most dyes are organic, and many organics are essentially transparent in the near infrared that's emitted by the LED.

I'm not clear what you thought should have happened. You place a transparent liquid in a transparent tube, you should still expect a transparent object.

Aside from that, measuring time of travel can be highly affected by how the leading and trailing boundaries of the liquid behave during actual flow conditions.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[ferragiov]

@IRstuff
If I place a semi-transparent liquid in a semi-transparent tube I would expect a semi-transparent object. What puzzles me is that tube+liquid is more trasparent than both tube only and no tube at all.

 

[IRstuff]

As the others have already indicated, it's focussing.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[btrueblood]

Either as further explanation or a completely different one, putting a fluid (with a refractive index closer to the tube's than air) inside the tube reduces the internal reflection of the LED light off of the inner wall. Try with a visible LED - I'd bet your clear poly tube acts as a light pipe with air inside (light refracts down the tube wall) and does it less with liquid on the i.d.

 

[controlsdude]

Might be the concept instead of using a send receive pe
But sticking with the PE concept, have you tried using a polarized lense or is this an option with this Omron product?
have you tried diffrent covers for the send receive like slits vertical or horizontal?

Different sensor
use a different type of sensor like a radar type and looking at the doppler effect then you would get a phase shift and use this for velocity.

 

[ferragiov]

Thanks everyone for their answers!
Unfortunatly I don't have visible-light coupled LED-sensors so at the moment I can't try this. The other switches I have come with an integrated comparator, so maybe I won't be able to see any difference between no-tube and tube-with-liquid comparator. The refractive index of the tube is 1.54 (PVC), water is 1.3. Both hypothesis (focusing and "fiber-optic" effect) seem to be reasonable, unfortunately as by now I can't try anything meaningful to sort things out. I attach the output of the phototransistor, which has a 1k resistor connected to the emitter. X axis is time in seconds (fluid runs at 100um/sec set by a siringe pump) and Y is volts. The initial 2 V refers to switch obstructed partially by the empty PVC tube (unobstructed gives 2.3 V). When the water approaches the sensor there is a current decrease, then it increases as the water covers the entire slit, until it stabilizes at about 3V. These values differ from the opening post since I slightly changed the setup. Maybe something can be inferred from the shape of this curve
Thanks again for everyone's answer!

 

[ferragiov]

PS sorry for the horrible x-axis. Each time step is 5 milliseconds

 

[OperaHouse]

The LED light goes out at an angle. Any liquid will focus it better. Our lab used to have a family day. We would set up a screen with a green plastic garbage bag and another one with window glass. The kids loved that you could see through the green bag and not the clear glass. Our best window was a 1 inch thick salt crystal. Pricey piece of rock salt. We had to keep it in a vacuum chamber to prevent it from frosting.

 

[OperaHouse]

Left out that was done with an IR camera.

 

[IRstuff]

OK, so, you have a demonstrable transition. What more do you need?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[MikeHalloran]

He was clearly hoping for a monotonic transition, with no dips or peaks, so he could use a comparator to get a precise and unequivocal switchpoint.

Right now, he's got a cylindrical sight glass, which has a dark line between the clear and less clear regions, or the regions where a scale placed behind the tube will be magnified differently depending on whether the lumen contains air or liquid.

You might think he'd get a little closer with the tube squashed flat, so all of the optical surfaces have zero curvature. ... but there will still be a dark meniscus, and the clear and less clear signals will be affected only by optical density, not refractive index, so he'd be worse off.

It's not a linear problem, so it doesn't matter if your photodiode is linear across five decades (as an old analog artificer bragged to me) or not.

The only way I know of to get a sharp repeatable transition _in_volume_ is to make the lumen at the detectors as small as is reasonably practical, so the meniscus passes through very rapidly even for low flow, and round, so the clear and less clear signals (again, really different magnifications) are more distinct from each other.












Mike Halloran
Pembroke Pines, FL, USA